Robust yield prediction of various farm processing unit

Problem Statement:

A new fast food chain is seeing rapid expansion over the past couple of years. They are now trying to optimize their supply chain to ensure that there are no shortages of ingredients. For this, they’ve tasked the data science team to come up with a model that could predict the output of each food processing farm over the next few years. These predictions could further increase the efficiency of their current supply chain management systems.

Loading Libraries

Reading Data and Basic understanding

Insight1

Insight2

Insight3

Insight4

Analyzing high and low yielded farms

  1. Converting date to datetime datatype and Extracting timestamp
  2. EDA
    1. Analysis of Top Yeilded Farm
    2. Analysis of Low Yeilded Farm
    3. Analysis of ingredient types

Finding high and low yielded farms

Analysis of Top Yielded Farm

Analysis of Low Yielded Field

Analysis of ingredient types

Merging the data

Merging train data

Merging test data

Univariate Analysis

Approach-1

ARMA

SARIMAX

FB Prophet

LSTM

Approach-2

Extracting Farm_ids

Making DataFrames for stationary and non Stationary

Fixing models and params

Making paramters dataframe

For Finding optimum params

SARIMA Forecast for 2017

  1. Model Running for batch 100
  2. List item

Creating Submissions for Sarimax

Multivariate Analysis

Approach-3

For ing_z

Forecasting ing_z for 2017

For ing_w